No abstract

The present study's two main purposes were to determine whether (a) training children to compare different parts of text improves detection of text errors and (b) self-controlled training of comparison produces more durable use of the strategy. Third graders heard expository passages, some containing explicit errors, and were asked to judge passage sensibility. Children who were taught to use a self-instructional routine specifying comparison of the two most recently presented sentences with each other and with the rest of the passage monitored comprehension immediately following training and 1 week later better than did subjects given minimal training. Teaching the two types of comparison without self-instruction produced only short-term benefits relative to minimal training alternatives. The results are consistent with Markman's (1979) coactivation hypothesis and with metacognitive theoretical assumptions about how to produce durable strategy use.

Children in Grades 3 and 4 were presented a referential communication task -generating a clue for an imaginary listener that would allow the listener to know which of two similar words (e.g., baby-child) was the referent. The optimal strategy for the task involves (a) generating an associate to the referent, (b) comparing the candidate clue with the referent and nonreferent, and (c) evaluating whether the clue is more highly associated with the referent than with the nonreferent. Children do not spontaneously carry out the comparison and evaluation components of this optimal strategy. Participants in the present complete instruction condition were taught to both compare and evaluate; their performance exceeded control performance for both easy and difficult materials. Comparison or evaluation training alone produced better clue production only for easy materials. The results are relevant to theoretical claims about referential communication and to the hypothesis that strategy instruction with grade school children should be elaborated and complete.Children's referential communication skills have been studied in detail in the last two decades (e.g., Asher, 1979;Asher & Wigfield, 1981). In a typical study, a child confronts an array of similar items -nonsense figures, colored geometric shapes, or words. The task is to provide clues to enable another person to decide which of the objects is the referent. Careful analyses of referential communications have suggested a potentially potent strategy for generating informative clues (e.g., Higgins, Fondacaro, & McCann, 1981;Rosenberg & Cohen, 1966):1. Communicators should generate an associate to the referent. 2. They should then compare the candidate clue with the referent and nonreferents to determine the extent of association to the two types of items.3. The communicators should evaluate whether a candidate clue has greater association with the referent than with any of the nonreferents.Although children probably generate clues by associating to the referent, they fail to compare and evaluate completely. Asher (1979) reviewed the relevant data.Asher and Wigfield (1981) studied the effects of training children to evaluate clues. They presented word pairs to third-and fourth-grade children (e.g., "piano-violin") and asked them to communicate to an imaginary listener which of the two words was the underlined referent. Asher and Wigfield's subjects viewed a film model who produced clues with the evaluation component emphasized in modeling. The child and model alternately produced clues for a total of 20 training pairs. The children were provided feedback on the adequacy of their clues, and over the course of practice trials modeling was reduced to covert clue evaluation with children encouraged to do their evaluations silently as well. Trained children provided better clues on the posttraining communication task than did control children who were exposed to the practice pairs but given no training, although positive effects

No abstract

No abstract